Construct Validation and Reliability Assessment of Tamil Version of Epworth Sleepiness Scale to Evaluate Daytime Sleepiness among Adolescents in Rural Puducherry, South India

INTRODUCTION

Excessive daytime sleepiness (EDS) can be defined as the increased propensity to sleep under circumstances considered inappropriate by the affected person and others. It is common symptom among various sleep disorders including sleep-related breathing disturbances such as obstructive sleep apnea.[123] Hence, to assess the level of general daytime sleepiness and identify sleep-related disorders in the general population, Johns developed a scale called Epworth sleepiness scale (ESS) in 1991.[4] This scale has been widely used in many countries throughout the world for research and clinical purposes.[5678] It is commonly used as it has simple structure and is easy to comprehend and complete rapidly.

ESS consists of questions related to eight distinctive situations in daily life and probability of falling asleep during each of these circumstances on a scale ranging from zero to three. In total, the score range from zero (normal) to 24 (very sleepy).[4] However, ESS was first developed for English population in Western countries. Hence, direct translation of the scale to other languages has certain limitations because of the economic and cultural differences. Several studies were conducted around the world for the translation and validation of the scale in different languages such as German, Turkish, Korean, Arabic, Urdu, Portuguese, Chinese, Greek, and Italian.[589101112131415] However, most of the studies were carried out for validation of questionnaire among adult population. Fewer attempts have been made around the world to validate the questionnaire among adolescents.[13]

Studies around India have shown that there has been a transition in sleep time of adolescents in both urban and rural setting, leading to poor sleep quality, daytime sleepiness, and poor academic performance. Main reason reported was the increasing use of mobile phones or social media and academic stress.[161718] Hence, there is a need to assess the sleep pattern among adolescents as they are becoming vulnerable to poor sleep pattern nowadays. However, there is no tool validated to use in local context. Hence, there is a need to adapt the western style questionnaire and validate the same for the local languages in India. ESS was validated for use in Hindi language.[19] However, we could not find any study on validation of ESS in any of the South Indian languages as Hindi version cannot be applied in South India, especially in rural setting because of lingual and cultural differences. Hence, the current study was carried out for translation of ESS into Tamil and then to validate the construct and assessment of reliability of the Tamil version of ESS among Tamil-speaking adolescents in Puducherry, South India.

MATERIALS AND METHODS

RESULTS

Sociodemographic characteristics

The mean age of the study participants was 14.1 (2.4) years. Table 1 explains the sociodemographic characteristics. Majority 425 (53.9%) participants belonged to early adolescent age group;[23] more than half 440 (55.8%) were boys; one-third 263 (33.3%) had completed secondary education; only one did not have any formal education. In total, 779 (98.7%) were students with almost three-fourth 575 (72.9%) studying in government school.

Table 1 Sociodemographic characteristics of the study participants (n=789)

Exploratory and confirmatory factor analysis

Before conducting EFA, Bartlett test of sphericity and KMO measures of sample adequacy were assessed to check the intercorrelation level of the items in ESS. KMO value 0.80 indicated that there was very good intercorrelation between the items of ESS, and Bartlett test showed high significance (Chi-square 2086.97, P < 0.001) which indicates that the results were acceptable.

Table 2 details the EFA conducted with principal component method with correlation matrix. Two factors (Factor 1 and 2) which had Eigenvalue of 3.49 and 1.09 were retained and rotated factor loading was generated using varimax rotation. Three items included in Factor 1 (Item 4, Item 5, and Item 7) explained 28.8% of variance. These three items focused mainly on excessive sleepiness during inactive period. Five items included in Factor 2 (Item 1, Item 2, Item 3, Item 6, and Item 8) explained 57.4% of variance. Items in Factor 2 focused mainly on excessive sleepiness during active period. The correlation between Factor 1 and Factor 2 was found to be 0.44.

Table 2 Results from exploratory factor analysis conducted through principal component method with varimax rotation on 8-point Epworth sleepiness scale questionnaire (n=789)

Two-factor models yielded through EFA were further analyzed through CFA. Initially, two-factor models were generated through structural equation modeling as shown in Figure 1. CFA revealed that two-factor model had Chi-square value of 215.57 with P < 0.001. Goodness-of-fit revealed acceptable CFIs (0.91) and SRMR (0.05) but mediocre RMSEA (0.10).

Close

Figure 1

Evaluated model of Epworth sleepiness scale structure generated through structural equation modeling. L1: Latent variable 1 (Factor 1), L2: Latent variable 2 (Factor 2), ep1–ep8: Epworth sleepiness scale item 1–8

Export to PPT

Reliability of Epworth sleepiness scale

The reliability coefficient (Cronbach's alpha) for the overall questionnaire was found to be 0.81 indicating very good internal consistency which was explained in Table 3. In addition, further factor-based analysis showed that in the first factor (3 items), Cronbach's reliability coefficient was 0.83, and in the second factor (5 items), Cronbach's reliability coefficient was 0.75.

Table 3 Internal consistency for the total Epworth sleepiness scale and two-factor model (n=789)

DISCUSSION

Tamil is the official and native language of the state Tamil Nadu and union territory Puducherry in India. There are approximately 70 million native speakers of Tamil in India. Outside India, Tamil is also official language in Singapore and Sri Lanka, and significant number of people in Malaysia is Tamil speakers. Validation of ESS and assessment of its reliability in Tamil can help the health professionals in clinical and research purposes. Hence, the main aim and purpose of the study were to validate the construct of ESS questionnaire and assess its reliability among Tamil-speaking adolescents.

Construct validation was done through exploratory factor and confirmatory factor method. First factor had three items and second factor had five items. All the factor loadings had high and statistically significant values. Further, CFA to compare and evaluate the goodness-of-fit of two-factor structure showed acceptable CFI and SRMR. The reliability assessment showed that the scale had good internal consistency similar to the English form of questionnaire, despite the lingual and cultural difference.[4] This reliability was also similar to the studies conducted in other countries among children and adolescents.[2425] Both the factor models also showed good internal consistency with acceptable Cronbach's alpha coefficient.

However, there were few problem items in the scale as discussed by a panel of experts. Certain questions such as riding a car were not culturally appropriate for adolescents, especially in a rural setting. We have rephrased the question as riding a bus instead of riding a car to make it more applicable to our study setting. Similar modifications were done by the previous study done in India for validation of the scale in Hindi language, in which they have removed such problem items and added more culturally appropriate questions, such as riding a bus or visiting a temple.[19] Hence, further modification can be done by discussing with the experts and community members to make the scale more suitable for the adolescents in Indian setting.

This study is the first one to validate the use of ESS questionnaire among adolescents in Indian population. Large sample size of 789 adolescents adds strength to the study. Homogeneity among the study population in the social, cultural, and economic context helps in generalization of the study results among the adolescents. However, validation done in our study can be applicable only to healthy Tamil-speaking adolescents for screening purpose and not applicable for adolescents with breathing-related sleep disturbances. We could not assess the test and retest reliability as the data collection was done at single time only. Content validity of the questionnaire cannot be tested in our study as the standard measure to diagnose daytime sleepiness by specialists was not done. Hence, further research on external validity of the questionnaire can be done to obtain the cutoff points for adolescents and check for comparability to the cutoff points of adults.

Daytime sleepiness assessment can be used to identify the population at risk of developing sleep disorders, especially the ones with breathing disturbances. Objective measurement of sleep quality through tests such as maintenance of wakefulness test and sleep latency test can be time-consuming and labor-intensive and not practically feasible for people in rural areas. In addition, the above-mentioned tests evaluate the patient in a specific circumstance conducted inside sleep laboratory and do not take the daily life activities into consideration. ESS can be used for identification of target population as it is easy to apply and understand and designed in such a way to reflect the average propensity of sleep during recent period.[10] It can even be applied at the routine adolescent clinics in primary health center or during school health visits by health workers.

Translation of this questionnaire in Tamil language and validation of it have provided a scientific basis for applying this scale among Tamil-speaking adolescents to screen for daytime sleepiness. However, this scale cannot be used as a diagnostic tool for distinguishing sleep disorders but used for screening purpose alone. Further modification can be done to the questionnaire to make it more culturally appropriate.

CONCLUSION

Our study concludes that the ESS for daytime sleepiness is internally valid and consistent with good reliability coefficient for application in Tamil-speaking rural adolescents. Construct validity assessment revealed a two-factor structure and acceptable fit indices for CFIs and SRMR. However, further research is required to prove the construct validity of an instrument as it is a continuous process of evaluation, reevaluation, refinement, and development.